Assessing the use of subgrid land model output to study impacts of land cover change

Subgrid information from land models has the potential to be a powerful tool for investigating land-atmosphere interactions, but relatively few studies have attempted to exploit subgrid output. In this study, we modify the configuration of the Community Land Model version CLM4.5 so that each plant functional type (PFT) is assigned its own soil column. We compare subgrid and grid cell-averaged air temperature and surface energy fluxes from this modified case (PFTCOL) to a case with the default configuration-a shared soil column for all PFTs (CTRL)-and examine the difference in simulated surface air temperature between grass and tree PFTs within the same grid cells (ΔTGT). The magnitude and spatial patterns of ΔTGT from PFTCOL agree more closely with observations, ranging from −1.5 K in boreal regions to +0.6 K in the tropics. We find that the column configuration has a large effect on PFT-level energy fluxes. In the CTRL configuration, the PFT-level annual mean ground heat flux (G) differs substantially from zero. For example, at a typical tropical grid cell, the annual G is 31.8 W m⁻² for the tree PFTs and −14.7 W m⁻² for grass PFTs. In PFTCOL, G is always close to zero. These results suggest that care must be taken when assessing local land cover change impacts with subgrid information. For models with PFTs on separate columns, it may be possible to isolate the differences in land surface fluxes between vegetation types that would be associated with land cover change from other climate forcings and feedbacks in climate model simulations.